Methodology

1. Ingestion

We start from the official CMS NPPES Data Dissemination file — currently the June 2026 vintage. The full file is parsed and loaded into a columnar store optimized for fast, read-only analytical queries across millions of records.

2. Normalization

• Names and addresses are cleaned and cased consistently for display.
• Provider entity type is split into individuals (Type 1) and organizations (Type 2).
• Free-text credential values are normalized (uppercased, punctuation stripped) so common credentials like MD, DO, and NP group cleanly.
• Practice state and city are standardized for facets and directory pages.

3. Taxonomy mapping

Each provider’s primary taxonomy code is joined to the NUCC Health Care Provider Taxonomy crosswalk to produce a human-readable specialty (grouping → classification → specialization).

4. Aggregation

We precompute counts by state, specialty, city, and credential so directory and facet pages load instantly without scanning the full corpus on every request.

5. Refresh

NPPES publishes a full monthly file with weekly incrementals. When we refresh, the entire dataset is rebuilt and every page’s “as of” date updates to reflect the new vintage.

Corrections

NPPES is the system of record. The fastest way to fix your information is to update it in NPPES; the change flows here on the next refresh. See corrections & disputes.